Is Passive Ownership Bigger than Estimated?

The Passive-Ownership Share is Double What You Think It Is

• Alex Chinco and Marco Sammon
• SSRN Working Paper
• A version of this paper can be found here
• Want to read our summaries of academic finance papers? Check out our Academic Research Insight category.

What are the research questions?

1. How much of the US stock market do passive investors own?
2. What accounts for the difference between published numbers and this study?
3. What does this research say about portfolio construction, if anything?

What are the Academic Insights?

1. The authors of this study develop and present a methodology for estimating the passive portion of the US equity market. Generally speaking, the increased volume that occurs at the close on reconstitution day, due to additions and deletions to the index, is converted to the total amount of capital tied to a particular index. This method captures strict end-of-day indexers and indicates that the generally accepted published value could be off by a factor of 2. For the most recent period, the authors estimate the passive portion is closer to 37.8%. Now compare that estimate to the value of 15%, published by the Investment Company Institute (ICI) as of the end of 2020.  The ICI calculates the sum of index mutual funds and ETF ownership of the US equity market, as an estimate of passively managed funds. The results for 2011 to 2021 are presented in Figure 1 below.  Note that the estimate hovers around 2x the ICI value across the years examined. The implication: Index funds and ETFs only account for at least 50% of all passively managed funds over the recent decade.
2. Here’s the caveat: End-of-day indexers are not captured in the ICI number. However, using this additional source of passive investing, the authors are able to use the count of ownership of end-of-day indexers to fill in potential gaps (more like craters?) in coverage.  End-of-day indexers are strictly passive investors who rebalance to their assigned benchmark at the end of every trading day. Pension funds and other institutional investors likely make up the difference. In this case, the subset of passive investors included here are tracking either the S&P500 or Russell 1000 or 2000.  Given there are only 3 indexes examined, the 37.8% is really a lower bound on the estimate of passive ownership in the US equity market. Taking that argument a bit further, a portion of passive investors rebalance at a much more leisurely pace and are not limited to end-of-day constraints. Again, the authors do not include that group in the 37.8%.
3. If the numbers presented in this research are more accurate by a factor of 2x, then the impact on modeling portfolio holdings could become a real issue. If investors are trying to determine what securities are held, why they are held, and who holds them in a portfolio, then index membership becomes has an order of magnitude of explanatory power exceeding that of our well-known pricing models. Not a surprising result, given that risk measures for client portfolios, are lower if the universe of stocks consists only of stocks with membership.  Although substitute stocks can be substituted by most risk models (MSCI-Barra, Northfield, etc.), if they are used, the estimate of tracking error will always be larger, depending on the number of stocks that have similar risk characteristics but are not members of the targeted index.

Why does it matter?

To the degree that policy decisions made around the degree of informativeness of market prices plus investor welfare are dependent on empirical measures of the magnitude of passive investing, this research argues for caution.  If the size of the passive market is understated by half in calculations, then the models used to inform policymakers need revision. Is the information content of market prices sufficient to guarantee the interests of market participants are well-served?

Why is it that no one else noticed this before? The approach presented here may be the wrong way to think about passive investing, or the theoretical models we have are simply incomplete but not incorrect.

A final thought from the authors:

Regardless of which interpretation you prefer, new methods are clearly needed when it comes to modeling the rise of passive investing. Existing models are not precise enough to recognize that the US passive-ownership share was off by a factor of two. The size of this blind spot poses a real problem for anyone trying to use these models to make policy decisions.

The most important chart from the paper

Abstract

We estimate that passive investors held at least 37.8% of the US stock market in 2020. This estimate is based on the closing volumes of index additions and deletions on reconstitution days. 37.8% is more than double the widely accepted previous value of 15%, which represents the combined holdings of all index funds. What’s more, 37.8% is a lower bound. The true passive-ownership share for the US stock market must be higher. This result suggests that index membership is the single most important consideration when modeling investors’ portfolio choice. In addition, existing models studying the rise of passive investing give no hint that prior estimates for the passive-ownership share were 50% too small. The size of this oversight restricts how useful these models can be for policymakers.